{"id":"https://openalex.org/W4412888953","doi":"https://doi.org/10.18653/v1/2025.findings-acl.4","title":"Diversifying the Expert Knowledge for Task-Agnostic Pruning in Sparse Mixture-of-Experts","display_name":"Diversifying the Expert Knowledge for Task-Agnostic Pruning in Sparse Mixture-of-Experts","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888953","doi":"https://doi.org/10.18653/v1/2025.findings-acl.4"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.4","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.4","pdf_url":"https://aclanthology.org/2025.findings-acl.4.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.4.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101802189","display_name":"Zeliang Zhang","orcid":"https://orcid.org/0000-0003-0329-8693"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zeliang Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100374771","display_name":"Xiaodong Liu","orcid":"https://orcid.org/0000-0001-8652-9818"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaodong Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101454917","display_name":"Hao Cheng","orcid":"https://orcid.org/0000-0002-0610-034X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao Cheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064805926","display_name":"Chenliang Xu","orcid":"https://orcid.org/0000-0002-2183-822X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chenliang Xu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5114910293","display_name":"Jianfeng Gao","orcid":"https://orcid.org/0000-0002-5702-6143"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jianfeng Gao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":7.5492,"has_fulltext":true,"cited_by_count":4,"citation_normalized_percentile":{"value":0.97481526,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"86","last_page":"102"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.933899998664856,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.933899998664856,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12761","display_name":"Data Stream Mining Techniques","score":0.9330999851226807,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.6598684787750244},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6585922837257385},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.6382272839546204},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.49924540519714355},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.38916900753974915},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.3273750841617584},{"id":"https://openalex.org/keywords/data-science","display_name":"Data science","score":0.3273504972457886},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.15947672724723816}],"concepts":[{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6598684787750244},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6585922837257385},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6382272839546204},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.49924540519714355},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38916900753974915},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3273750841617584},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.3273504972457886},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.15947672724723816},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.4","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.4","pdf_url":"https://aclanthology.org/2025.findings-acl.4.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.4","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.4","pdf_url":"https://aclanthology.org/2025.findings-acl.4.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/2","score":0.47999998927116394,"display_name":"Zero hunger"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888953.pdf","grobid_xml":"https://content.openalex.org/works/W4412888953.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2373300491","https://openalex.org/W2395294869","https://openalex.org/W2378744544","https://openalex.org/W2594301978","https://openalex.org/W2379704676","https://openalex.org/W1998810860","https://openalex.org/W4206442282","https://openalex.org/W2384505857","https://openalex.org/W2355171581","https://openalex.org/W4229439743"],"abstract_inverted_index":{"In":[0],"this":[1],"work,":[2],"we":[3],"address":[4],"the":[5,27,61,66,102,106,112,118],"memory":[6,33],"overhead":[7],"of":[8,30,53,68,95,105],"deploying":[9],"Mixture-of-Experts":[10],"(MoE)":[11],"architectures":[12],"in":[13,131],"Large":[14],"Language":[15],"Models":[16],"(LLMs).While":[17],"MoE":[18,75],"layers":[19],"improve":[20,60],"LLM":[21],"performance":[22,120],"without":[23],"increasing":[24],"inference":[25],"costs,":[26],"evergrowing":[28],"number":[29],"experts":[31,43,58],"inflates":[32],"requirements,":[34],"hindering":[35],"practical":[36],"deployment.Our":[37],"empirical":[38],"study":[39],"reveals":[40],"that":[41,84],"some":[42],"encode":[44],"redundant":[45],"knowledge":[46],"during":[47],"pretraining.We":[48],"thus":[49],"propose":[50],"a":[51,93],"method":[52,70,86],"grouping":[54],"and":[55,80,108,117,125],"pruning":[56,72,90,103],"similar":[57],"to":[59],"model's":[62],"parameter":[63],"efficiency.We":[64],"validate":[65],"effectiveness":[67],"our":[69,85],"by":[71],"three":[73],"state-of-the-art":[74],"architectures,":[76],"including":[77],"Mixtral,":[78],"Deepseek-MoE,":[79],"Qwen.The":[81],"evaluation":[82],"shows":[83],"outperforms":[87],"other":[88],"model":[89],"methods":[91],"on":[92,111],"range":[94],"natural":[96],"language":[97],"tasks.":[98],"We":[99],"only":[100],"show":[101],"results":[104,127],"Deepseek":[107],"Qwen":[109],"models":[110],"most":[113],"challenging":[114],"MMLU":[115],"task":[116],"average":[119],"across":[121],"MMLU,":[122],"BoolQ,":[123],"OpenBookQA,":[124],"RTE.Full":[126],"can":[128],"be":[129],"found":[130],"supplementary.":[132]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":3}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}